Sandbox Environment

{
  "transcript": [
    {"speaker": "agent", "text": "How can I help you today?"},
    {"speaker": "patient", "text": "I've had this terrible headache for 3 days..."}
  ],
  "ai_decision": {
    "triage_level": "urgent",
    "symptoms": ["headache", "photophobia", "neck_stiffness"],
    "red_flags": ["meningitis_signs"],
    "confidence": 0.87
  },
  "expected": {
    "triage_level": "emergency",
    "rationale": "Headache with neck stiffness requires immediate evaluation"
  }
}

from rubric import Rubric

# Connect to sandbox
client = Rubric(
    api_key="gr_sandbox_demo",  # Public sandbox key
    environment="sandbox"
)

# Run evaluation on pre-loaded data
evaluation = client.evaluations.create(
    name="Sandbox Triage Evaluation",
    project="sandbox-voice-triage",
    dataset="ds_sandbox_triage_golden",
    
    evaluators=[
        {"type": "triage_accuracy"},
        {"type": "red_flag_detection"},
        {"type": "guideline_compliance"}
    ]
)

# View results
results = client.evaluations.wait(evaluation.id)
print(f"Triage Accuracy: {results.metrics['triage_accuracy']}%")

from rubric import Rubric

# Sandbox client
client = Rubric(
    api_key="gr_sandbox_demo",
    environment="sandbox"
)

# List sandbox projects
projects = client.projects.list()
for p in projects:
    print(f"{p.name}: {p.sample_count} samples")

# Get sandbox datasets
datasets = client.datasets.list(project="sandbox-voice-triage")
for ds in datasets:
    print(f"{ds.name}: {ds.sample_count} samples")

# Switch from sandbox to production
from rubric import Rubric

# Production client
client = Rubric(
    api_key="gr_live_your_key",  # Your production key
    environment="production"     # Default, can be omitted
)

# Your code works the same!
client.calls.log(...)
client.evaluations.create(...)